@article {bnh-8361, title = {Determining threshold conditions for extreme fire behaviour - final project report}, year = {2022}, month = {09/2022}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

In the last decade, there have been extreme wildfire events around the world resulting in substantial social, economic and environmental impacts. They threaten many lives and cost billions of dollars in damage. Climate change is making the fire seasons around the world even worse by extending the number of dry and hot days [1-4]. A longer fire season is expected to result in more frequent and severe fires [5, 6]. Australia{\textquoteright}s bushfire season 2019/20 (Black Summer fires hereafter) appears to have supported these conclusions in terms of the ecological consequences and impacts on human populations [7].

In most cases, these consequences are the result of dynamic fire behaviours (DFBs) [8-11]. The DFBs are localised dynamic events that occur within fires, whereby physical feedbacks greatly enhance fire intensities and rates of spread. Understanding and having the ability to predict DFBs in wildfire events is essential to ensure the safety of communities living in or near the Wildland-Urban Interface.

In this regard, the project {\textquoteleft}Determining threshold conditions for extreme fire behaviour{\textquoteright} was focused on the understanding and analysis of dynamic fire effects; their influence on fire behaviour and structures; and the potential of including these effects in fire behaviour models and new building standards.

}, issn = {735}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-7499, title = {Determining threshold conditions for extreme fire behaviour - annual report 2019-2020}, number = {626}, year = {2020}, month = {11/2020}, institution = {Bushfire and Natural Hazards CRC}, address = {MELBOURNE}, abstract = {

At this phase the project was focused on development of a new method to test flammability of live vegetation in dynamic conditions and understanding influence of climatic changes on the 2019/20 bushfire season in New South Wales (NSW), Victoria, and South Australia (SA).

Understanding live vegetation fuel properties and how they behave when exposed to radiant heat and flame allows us to better predict fire behaviour in forested areas. This study aims to determine a more effective, replicable and accurate method of testing flammability in live vegetation by comparing the impact different radiant heating regimes have on the ignitibility of live vegetation samples. Current methodologies are limited in their ability to provide accurate quantification of flammability due to their reliance on static heat flux exposure, which does not accurately replicate how live plants experience radiative heat flux during a wildfire in their natural environment. Two heating regimes were tested for this study {\textendash} a static heat flux to reflect current methods and a dynamic (increasing) heat flux to more accurately replicate real conditions of an approaching fire front. Piloted-ignition and unpiloted-ignition were also tested for both of these heating regimes. A Variable Heat Flux (VHFlux) Apparatus was used to study flammability of Acacia floribunda, Cassinia arcuata, Pinus radiata and bark from Eucalyptus obliqua. Time to pyrolysis (production of volatile products), smouldering, flaming ignition, complete consumption and radiant exposure (the radiant energy received by a sample over a time of heating, He) were used as ignitability measures. It was observed that time and radiant exposure required to reach flaming ignition (and the other ignitibility metrics) was higher under a dynamic heating regime. It was also observed that the presence of a pilot igniter greatly increased the number of samples that reached flaming ignition, and decreased the time and He required to reach flaming ignition (and the other ignitibility metrics). These results suggest clear differences observed between heating regimes for time and Herequired for ignition and other ignitibility measures, which supports the validity of using dynamic heating regimes and the VHFlux apparatus as a standardised method. Adoption of this methodology is recommended to ensure more realistic data on flammability of individual plant species and plant communities, which will ultimately lead to better informed and more accurate wildfire behaviour modelling.

There is no doubt that the fire season of 2019/20 was extraordinary. A total of 18,983,588hectares were burned, 3113 houses and 33 lives lost in 15,344 bushfires in Black Summer fires. NSW had the highest number of fires, area burned, houses and lives lost for the last 20 years. Two mega-blazes occurred in NSW and burned more than in any fire season during the last 20 years. Victoria had the highest number of fires, area burned, and houses lost (except for the Black Saturday fires). SA had the highest number of houses lost in the last 20 years. Relationships between the burned area and number of fires, the houses and lives lost had positive trend for all states irrespective of the dataset. A negative relationship between the houses and lives lost for SA was the only exception. Multiple studies show that fire weather will become more severe in many regions around the world. Based on this and observed positive trends for all categories for NSW and Victoria, it is likely that the values will continue to increase in these states in the future. SA before 2019/20 was in a relatively good position showing negative trends for almost all categories. However, the 2019/20 fire season changed that for the worse. The magnitude of effect from increased fire weather may depend on how these conditions alter vegetation across Australia, however the indications shown in this analysis are concerning for fire managers.

Smoke from bushfires significantly impacted on people with cardiovascular and respiratory problems and increased mortality. It also had indirect impact on the economy by disrupting communities. The total impact of the 2019/20 bushfire season to the economy is estimated to be as much as A$40 billion. Due to the record burned area, at least 1 billion vertebrate animals were lost. It will take many years to restore the economy in impacted areas, and for animal and vegetation biodiversity to recover. Understanding of high-level trends of number of fires, area burned, houses and lives lost for the last two decades in south-eastern Australia will provide useful insights to fire managers for future strategies and policies.

}, keywords = {Behaviour, conditions, extreme fire, threshold}, issn = {626}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-7343, title = {Effect of weather forecast errors on fire growth model projections}, journal = {International Journal of Wildland Fire}, year = {2020}, month = {08/2020}, abstract = {

Fire management agencies use fire behaviour simulation tools to predict the potential spread of a fire in both risk planning and operationally during wildfires. These models are generally based on underlying empirical or quasi-empirical relations and rarely are uncertainties considered. Little attention has been given to the quality of the input data used during operational fire predictions. We examined the extent to which error in weather forecasts can affect fire simulation results. The study was conducted using data representing the State of Victoria in south-eastern Australia, including grassland and forest conditions. Two fire simulator software packages were used to compare fire growth under observed and forecast weather. We found that error in the weather forecast data significantly altered the predicted size and location of fires. Large errors in wind speed and temperature resulted in an overprediction of fire size, whereas large errors in wind direction resulted in an increased spatial error in the fire{\textquoteright}s location. As the fire weather intensified, fire predictions using forecast weather under predicted fire size, potentially resulting in greater risks to the community. These results highlight the importance of on-ground intelligence during wildfires and the use of ensembles to improve operational fire predictions.

}, keywords = {Bayesian Network, fire prediction, meteorological forecast, sensitivity, simulation}, doi = {https://doi.org/10.1071/WF19199}, url = {https://www.publish.csiro.au/wf/wf19199}, author = {Trent Penman and Dan Ababei and Jane Cawson and Brett Cirulis and Thomas Duff and Swedosh, W and James Hilton} } @article {bnh-7475, title = {Exploring the key drivers of forest flammability in wet eucalypt forests using expert-derived conceptual models}, journal = {Landscape Ecology}, volume = {35}, year = {2020}, month = {06/2020}, pages = {1775{\textendash}1798}, abstract = {

Context

Fire behaviour research has largely focused on dry ecosystems that burn frequently, with far less attention on wetter forests. Yet, the impacts of fire in wet forests can be high and therefore understanding the drivers of fire in these\ systems\ is vital.

Objectives

We sought to identify and rank by importance the factors plausibly driving flammability in wet eucalypt forests, and describe relationships between them. In doing so, we formulated a set of research priorities.

Methods

Conceptual models of forest flammability in wet eucalypt forests were elicited from 21 fire experts using a combination of elicitation techniques. Forest flammability was defined using fire occurrence and fireline intensity as measures of ignitability and heat release rate, respectively.

Results

There were shared and divergent opinions about the drivers of flammability in wet eucalypt forests. Widely agreed factors were drought, dead fine fuel moisture content, weather and topography. These factors all influence the availability of biomass to burn, albeit their effects and interactions on various dimensions of flammability are poorly understood. Differences between the models related to lesser understood factors (e.g. live and coarse fuel moisture, plant traits, heatwaves) and the links between factors.

Conclusions

By documenting alternative conceptual models, we made shared and divergent opinions explicit about flammability in wet forests. We identified four priority research areas: (1) quantifying drought and fuel moisture thresholds for fire occurrence and intensity, (2) modelling microclimate in dense vegetation and rugged terrain, (3) determining the attributes of live vegetation that influence forest flammability, (4) evaluating fire management strategies.

}, keywords = {Cognitive mapping, Conceptual models, Expert elicitation, Fire behaviour, fire intensity, flammability, Structured decision-making, Structured expert judgement, Wet forest, Wildfire}, doi = {https://doi.org/10.1007/s10980-020-01055-z}, url = {https://link.springer.com/article/10.1007/s10980-020-01055-z}, author = {Jane Cawson and Victoria Hemming and Ackland, A and Wendy R. Anderson and David Bowman and Ross Bradstock and Brown, T and Jamie Burton and Geoffrey J. Cary and Thomas Duff and Alex Filkov and Furlaud, James M. and Tim Gazzard and Kilinc, Musa and Petter Nyman and Ross Peacock and Mike Ryan and Jason J. Sharples and Gary J. Sheridan and Tolhurst, K.G. and Tim Wells and Phil Zylstra and Trent Penman} } @article {bnh-5830, title = {Determining threshold conditions for extreme bushfire behaviour annual report 2018-2019}, number = {508}, year = {2019}, month = {09/2019}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

Crown fires in forest ecosystems can pose a major threat to life and property due to their high intensities and rapid rates of spread. However, research into the prediction of crown fire dynamics in the Eucalyptus forests of Australia is limited. Previous studies have focused on coarse temporal scales, utilised low resolution weather based predictors, and often disregard the spatial nature of crown fires. Our study aimed to use observations from large wildfires in eucalypt forests to develop an empirical model to predict the likelihood of crown fire events using environmental predictors at an hourly scale. Our study was conducted in south-eastern Australia using data from fifteen large wildfires that occurred between 2009 and 2015. Fire severity maps were created for each fire at a 30\ m resolution using Landsat imagery from which we calculated the proportion of 30 m pixels experiencing crown fire within a 150 x 150 m window (2.25 ha). Predictor variables were chosen to represent the four key environmental drivers of fire behaviour, namely fuel moisture (i.e. live and dead fuel), fuel load and structure (i.e. surface, elevated, and bark fuels, and tree height), fire weather (i.e. vapour-pressure deficit, wind speed, relative wind direction) and topography (i.e. slope and ruggedness). Random Forests were used to model the effect of environmental drivers on the proportion of crown fire. Fuel moisture content variables were the best predictors of probability of crown consumption. Topographic variables and fire weather had only an intermediate influence and fuel load and structure had the lowest influence. Crown fire runs largely occurred when thresholds in vapour-pressure deficit (\<4\ kPa) and dead fuel moisture content (\<7\%) were exceeded. Predictions from the model showed a high degree of agreement with the raw fire severity maps. The proposed models have the potential to provide guidance on the likelihood of crown fire during fire events.

}, keywords = {Bushfire, extreme bushfires, fire behaviour., weather conditions}, issn = {508}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-5424, title = {Determining Threshold Conditions for Extreme Fire Behaviour Annual Report 2017-2018}, number = {460}, year = {2019}, month = {03/2019}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

Extreme fires cause disproportionate impacts on the environment and the community. There are significant incentives to being able to predict their occurrence and behaviour. Most existing fire behaviour models have been developed based on data and observations of fires that were small to moderate in size. Consequently, they are not able to emulate the dynamic bushfire behaviour that can occur under extreme conditions.

The main aim of this project is to investigate the conditions and processes under which bushfire behaviour undergoes major transitions, including fire convection and plume dynamics, evaluating the consequences of eruptive fire behaviour (spotting events, convection driven wind damage, rapid fire spread) and determining the combination of conditions for such behaviours to occur (e.g. unstable atmosphere, fuel properties and weather conditions). To do this the project was separated into two phases. The first phase of the project was focused on data collection about extreme fires, analysis the frequency of occurrence of extreme fire phenomena and determination the potential of including them in fire behaviour models.

}, keywords = {extreme fire behaviour, fire behavour, fire management}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-6393, title = {Frequency of Dynamic Fire Behaviours in Australian Forest Environments}, journal = {Fire}, volume = {3}, year = {2019}, month = {12/2019}, abstract = {

Wildfires can result in significant social, environmental and economic losses. Fires in which dynamic fire behaviours (DFBs) occur contribute disproportionately to damage statistics. Little quantitative data on the frequency at which DFBs occur exists. To address this problem, we conducted a structured survey using staff from fire and land management agencies in Australia regarding their experiences with DFBs. Staff were asked which, if any, DFBs were observed within fires greater than 1000 ha from the period 2006{\textendash}2016 that they had experience with. They were also asked about the nature of evidence to support these observations. One hundred thirteen fires were identified. Eighty of them had between one and seven DFBs with 73\% (58 fires) having multiple types of DFBs. Most DFBs could commonly be identified through direct data, suggesting an empirical analysis of these phenomena should be possible. Spotting, crown fires and pyro-convective events were the most common DFBs (66\%); when combined with eruptive fires and conflagrations, these DFBs comprise 89\% of all cases with DFBs. Further research should be focused on these DFBs due to their high frequencies and the fact that quantitative data are likely to be available.

}, keywords = {anecdotal data, direct data, dynamic fire behaviours, indirect data, Wildfire}, doi = {https://doi.org/10.3390/fire3010001}, url = {https://www.mdpi.com/2571-6255/3/1/1}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-4998, title = {Determining threshold conditions for extreme fire behaviour}, number = {417}, year = {2018}, month = {10/2018}, institution = {Bushfire and Natural Hazards CRC}, abstract = {

Flame spread is an important process in the propagation of bushfires. The likelihood of ignition and combustion rates of fuels are dependent on the type and nature of the heat flux. The majority of previous research has used static heat flux, whereby a consistent heating source is used to ignite samples in a laboratory setting. This is despite the highly dynamic heating regimes typically observed during structural and wildland fires.

}, keywords = {Fire, fire behaviour., fire modelling, fire severity, modelling}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @conference {bnh-4743, title = {Extreme fire behaviours: Surveying fire management staff to determine behaviour frequencies and importance}, booktitle = {AFAC18}, year = {2018}, month = {09/2018}, publisher = {Bushfire and Natural Hazards CRC}, organization = {Bushfire and Natural Hazards CRC}, address = {Perth}, abstract = {

An understanding how bushfires cause damage is important if they are to be effectively managed. Extreme fire behaviours (EFBs) are phenomena that occur within intense fires that have been shown to contribute greatly to their to impacts. However, there exists little understanding regarding how often particular EFBs occur, how these contribute to fire behaviour and what importance should be allocated to each in the development of models for decision support. To address this problem, we surveyed fire fighters from fire and land management agencies in Australia regarding their experiences with EFBs. All fires greater than 1000 ha in the period 2006-2016 were considered in the survey. Representatives were asked which, if any, EFBs they had observed and whether there was any documentation to support these observations. We found that EFBs are common in large fires. In more than 60 \% of case studies, each bushfire had two and more EFBs simultaneously (or one after another). Our survey indicated that Spotting, Crown fires, Pyro-convective events, Eruptive fires and Conflagrations are the most commonly observed EFBs, and so should be a priority for research. The relative commonness of direct evidence available for EFBs is indicative that there should be the potential for further study of these phenomena.\ 

}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-4458, title = {Improving Fire Behaviour Data Obtained from Wildfires}, journal = {Forests}, volume = {9}, year = {2018}, month = {02/2018}, abstract = {

Organisations that manage wildfires are expected to deliver scientifically defensible decisions. However, the limited availability of high quality data restricts the rate at which research can advance. The nature of wildfires contributes to this: they are infrequent, complex events, occur with limited notice and are of relatively short duration. Some information is typically collected during wildfires, however, it is often of limited quantity and may not be of an appropriate standard for research. Here we argue for a minimum standard of data collection from every wildfire event to enhance the advancement of fire behaviour research and make research findings more internationally relevant. First, we analyse the information routinely collected during fire events across Australia. Secondly, we review research methodologies that may be able to supplement existing data collection. Based on the results of these surveys, we develop a recommended list of variables for routine collection during wildfires. In a research field typified by scarce data, improved data collection standards and methodologies will enhance information quality and allow the advancement in the development of quality science.

}, keywords = {data collection and management, Fire behaviour, research utilization, standard.}, doi = {10.3390/f9020081}, url = {http://www.mdpi.com/1999-4907/9/2/81}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-5279, title = {A lightning-caused wildfire ignition forecasting model for operational use}, journal = {Agricultural and Forest Meteorology}, volume = {253-254}, year = {2018}, month = {05/2018}, pages = {16}, chapter = {233}, abstract = {

Lightning-caused wildfires are responsible for substantial losses of lives and property worldwide. Convective storms can create large numbers of ignitions that can overwhelm suppression efforts. Both long- and short-term risk planning could benefit from daily, spatially-explicit forecasts of lightning ignitions. We fitted a logistic regression generalised additive model to lightning-caused ignitions in the state of Victoria, Australia. We proposed a new method for model selection that complemented existing methods and further reduced the number of variables in the model with minimal change to predictive power. We introduced an approach for deconstructing ignition forecasts into contributions from the individual covariates, which could allow model output to be more readily integrated with existing intuitive understandings of ignition likelihood. Our method of model selection reduced the number of variables in the model by 37.5\% with little change to the predictive power. The final model showed good predictive ability (AUC 0.859) and we demonstrated the utility of the model for short term forecasting by comparing model predictions with observed lightning-caused fires over three time periods, two of which had extreme fire conditions, while the third was randomly chosen from our validation dataset. The model presented in this paper shows good predictive power and advancements in model output could allow fire managers to more easily interpret model forecasts.

}, doi = {10.1016/j.agrformet.2018.01.037}, url = {https://www.sciencedirect.com/science/article/pii/S0168192318300376}, author = {Nicholas Read and Thomas Duff and Peter Taylor} } @article {bnh-4206, title = {Determining threshold conditions for extreme fire behaviour: annual project report 2016-17}, number = {319}, year = {2017}, month = {09/2017}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

Organisations that manage bushfires are expected to deliver scientifically defensible decisions. However, the limited availability of high quality data restricts the rate at which research can advance. The nature of bushfires contributes to this; they are infrequent, complex events, occur with limited notice and are of relatively short duration. Some information is typically collected during bushfires however it may not be of an appropriate standard for research. In the past year we have focused on the information that is typically collected during fires. First we reviewed the information routinely collected during fire events across Australia. Secondly, we reviewed research methodologies that may be able to supplement existing data collection. Based on the results of these surveys, we developed a recommended list of attributes for routine collection during bushfires. We also suggest standards of data collection from bushfire events to enhance the advancement of fire behaviour research and make research findings more internationally relevant. In a research field typified by scarce data, improved data collection standards and methodologies will enhance information quality and allow the advancement in the development of quality science (1). In addition to the fire data review, we investigated embers, including their production and how they burn (2,3).

}, issn = {319}, author = {Alex Filkov and Thomas Duff and Trent Penman} } @article {bnh-2922, title = {Determining threshold conditions for extreme fire behaviour: Annual project report 2015-2016}, number = {175}, year = {2016}, month = {08/2016}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

As extreme fires cause a disproportionate amount of impact to the environment and the community, there are significant incentives to being able to predict their occurrence and behaviour. Most existing fire behaviour models have been developed based on data and observations of fires that were small to moderate in size. Consequently, they are not able to emulate the dynamic bushfire behaviour that can occur under extreme conditions.

Indeed, current operational fire spread models assume that fires will burn at an approximately constant (quasi-steady) rate of spread under a specific set of environmental conditions (e.g. VESTA, McArthur Mk5, CSIRO models). While a number of advances have been made in understanding bushfire development under extreme conditions, these have not been quantified in a manner that is suitable for inclusion in a fire behaviour modelling framework.

The main aims of this project are to investigate the conditions and processes under which bushfire behaviour undergoes major transitions, including fire convection and plume dynamics, evaluating the consequences of eruptive fire behaviour (spotting events, convection driven wind damage, rapid fire spread) and determining the combination of conditions for such behaviours to occur (e.g. unstable atmosphere, fuel properties and weather conditions). To do this the collation and analysis of existing data on extreme fire behaviour will be done.

}, issn = {175}, author = {Thomas Duff and Trent Penman and Alex Filkov} } @conference {bnh-2932, title = {Wind speed reduction induced by post-fire vegetation regrowth}, booktitle = {AFAC16}, year = {2016}, month = {08/2016}, publisher = {Bushfire and Natural Hazards CRC}, organization = {Bushfire and Natural Hazards CRC}, address = {Brisbane}, abstract = {

In the current suite of operational fire spread models, wind speeds measured in the open environment (above the vegetation layer) are modified to represent wind speeds at {\textquoteleft}mid-flame{\textquoteright} height using adjustment factors. In general, these adjustment factors assume constant vertical wind speed profiles throughout the vegetation layer. However, empirical studies have shown that wind speeds beneath canopies can vary significantly with height above ground as well as with forest type and prevailing wind speed. Empirical wind reduction profiles have been developed for a number of different forest types in flat terrain using data collected across Victoria, Australia.


The present research aims to extend these empirical studies to better understand the impacts of topography and post-fire vegetation regrowth on the reduction of wind speeds beneath the canopy. Wind data collected over fire affected regions of rugged terrain in South Eastern Australia are used to analyse wind speed reduction induced by post-fire regrowth and complex topography. A secondary study is used to analyse wind speed reduction caused by Radiata pine plantation in undulating terrain.


Results of this study suggest that empirical wind reduction profiles perform well at the broader landscape-scale, i.e. ridge tops and valley floors. However, more complex topographical features appear to have a compounding affect on wind speed reduction within rugged terrain. Through better understanding of wind speed reduction beneath the canopy across landscapes from mountainous ranges through to flat plains, wind speed reduction models for bushfire spread prediction can be adapted to incorporate the variation observed in vertical wind speed profiles within the vegetation layer.

}, author = {Rachael Quill and Kangmin Moon and Jason J. Sharples and Leesa Sidhu and Thomas Duff and Tolhurst, K.G.} } @article {bnh-2395, title = {Determining threshold conditions for extreme fire behaviour}, year = {2015}, author = {Thomas Duff and Trent Penman} } @article {bnh-2340, title = {Determining threshold conditions for extreme fire behaviour: Annual project report 2014-2015}, number = {133}, year = {2015}, month = {02/11/2015}, institution = {Bushfire and Natural Hazards CRC}, address = {Melbourne}, abstract = {

Fire behavior models that predict the progression of bushfires are becoming increasingly important in management.\  Most existing models have been developed based on data and observations of fires burning under relatively mild conditions.\  If the models are to be relied upon for fires that occur under extreme weather conditions (where fires are fast moving and more intense), these conditions must also be considered in their design.

There is increasing evidence that there are particular fire phenomena that occur only in extreme fire conditions.\  These include fire tornados, atmospheric coupling, ember storms and vorticity driven lateral spread.\  Research into such phenomena is limited, there is still much to learn about when they occur and what their effect is on fire behavior.\  Currently there are no operational fire spread models that can accommodate these important effects. To account for these phenomena, it is first necessary to describe them and the conditions under which they occur, including fuel conditions, surface weather and atmospheric profiles.

This project is designed to build on our knowledge of the unique features of extreme fires by a) Collating observations of extreme fires that have occurred in Australia in recent years, b) Analysing fire phenomena in conjunction with accessory information (ie. Weather, fuel and topography) and c) Developing mathematical relationships to describe important fire phenomena.

}, issn = {133}, author = {Thomas Duff and Trent Penman} } @article {bnh-1843, title = {Erratum to {\textquotedblleft}Using discrete event simulation cellular automata models to determine multi-mode travel times and routes of terrestrial suppression resources to wildland fires"}, journal = {European Journal of Operational Research}, volume = {245}, year = {2015}, month = {08/2015}, pages = {339-340}, chapter = {339}, abstract = {

Forest fires can impose substantial social, environmental and economic burdens on the communities on which they impact. Well managed and timely fire suppression can demonstrably reduce the area burnt and minimise consequent losses. In order to effectively coordinate emergency vehicles for fire suppression, it is important to have an understanding of the time that elapses between vehicle dispatch and arrival at a fire. Forest fires can occur in remote locations that are not necessarily directly accessible by road. Consequently estimations of vehicular travel time may need to consider both on and off road travel. We introduce and demonstrate a novel framework for estimating travel times and determining optimal travel routes for vehicles travelling from bases to forest fires where both on and off road travel may be necessary. A grid based, cost-distance approach was utilised, where a travel time surface was computed indicating travel time from the reported fire location. Times were calculated using a discrete event simulation cellular automata (CA) model, with the CA progressing outwards from the fire location. Optimal fastest travel paths were computed by recognising chains of parent{\textendash}child relationships. Our results achieved comparable results to traditional network analysis techniques when considering travel along roads; however the method was also demonstrated to be effective in estimating travel times and optimal routes in complex terrain.

}, doi = {doi:10.1016/j.ejor.2015.03.025}, url = {http://www.sciencedirect.com/science/article/pii/S0377221715002325}, author = {Thomas Duff and Chong, Derek and Tolhurst, K.G.} } @article {bnh-1841, title = {Using discrete event simulation cellular automata models to determine multi-mode travel times and routes of terrestrial suppression resources to wildland fires}, journal = {European Journal of Operational Research}, volume = {241}, year = {2015}, month = {03/2015}, pages = {763-770}, chapter = {763}, abstract = {

Forest fires can impose substantial social, environmental and economic burdens on the communities on which they impact. Well managed and timely fire suppression can demonstrably reduce the area burnt and minimise consequent losses. In order to effectively coordinate emergency vehicles for fire suppression, it is important to have an understanding of the time that elapses between vehicle dispatch and arrival at a fire. Forest fires can occur in remote locations that are not necessarily directly accessible by road. Consequently estimations of vehicular travel time may need to consider both on and off road travel. We introduce and demonstrate a novel framework for estimating travel times and determining optimal travel routes for vehicles travelling from bases to forest fires where both on and off road travel may be necessary. A grid based, cost-distance approach was utilised, where a travel time surface was computed indicating travel time from the reported fire location. Times were calculated using a discrete event simulation cellular automata (CA) model, with the CA progressing outwards from the fire location. Optimal fastest travel paths were computed by recognising chains of parent{\textendash}child relationships. Our results achieved comparable results to traditional network analysis techniques when considering travel along roads; however the method was also demonstrated to be effective in estimating travel times and optimal routes in complex terrain.

}, keywords = {Transport; Network; environment and climate change; Routing; Simulation}, doi = {doi:10.1016/j.ejor.2014.09.019}, url = {http://www.sciencedirect.com/science/article/pii/S0377221714007401}, author = {Thomas Duff and Chong, Derek and Tolhurst, K.G.} } @article {BF-4281, title = {Quantifying spatio-temporal differences between fire shapes: Estimating fire travel paths for the improvement of dynamic spread models}, journal = {Environmental Modelling \& Software}, volume = {46}, year = {2013}, month = {08/2013}, pages = {33-43}, chapter = {33}, abstract = {Dynamic fire spread models are a recent development in landscape management that provide for the simulation of the spread of fires through time under complex weather conditions. These allow risks to be assessed and resources to be strategically managed. The need for reliable and accurate fire models is of particular importance in the face of recent catastrophic wildfires in Australia, Europe and the United States. However, while fire spread models are developed using physical knowledge and empirical observations, there are few techniques which can be used to objectively assess the {\textquoteleft}goodness of fit{\textquoteright} of spatial predictions of fire spread. We propose a new method to allow the comparison of fire perimeters, providing for the discrimination of sources of simulation error and assisting in the collection of empirical spread data from observed fires. Differences between fire perimeters are quantified using linear vectors aligned with the direction of spread of the perimeter being sampled. These can provide an indication of difference in terms of the fire spread distance on the ground. The location, direction and length of these vectors can be used to assess spread rates to assist with model calibration. We demonstrated the utility of this method using a case study which assessed differences between the observed and simulated progression of an Australian wildfire. The new indices were found to be effective descriptors of differences in fire shape and hold potential for the spatial evaluation of fire spread models. The indices can be used to compare similar fire shapes; however they are unsuited for cases where there are large differences between perimeters.}, keywords = {Bushfire, Morphometrics, Perimeter, Shape analysis, Spatial validation, Wildfire}, doi = {http://dx.doi.org/10.1016/j.envsoft.2013.02.005}, url = {http://www.sciencedirect.com/science/article/pii/S1364815213000455}, author = {Thomas Duff and Chong, Derek and Tolhurst, K.G.} } @article {BF-4295, title = {Sensitivity Analysis of PHOENIX RapidFire}, year = {2013}, month = {05/2013}, abstract = {An analysis of the sensitivity of the outputs of PHOENIX Rapidfire (PHOENIX) to a range of inputs and simulation parameters was undertaken. This was done using two separate methods; assessment of model response in an artificially generated idealised landscape and assessment using case-studies of real fires. The ideal landscape was used to evaluate model sensitivity in response to temperature, relative humidity, wind speed, fuel type and wind direction relative to slope. The model was evaluated under two sets of weather conditions, mild (representing moderate fire spread potential) and extreme (representing high fire spread potential). Each scenario was evaluated for each of two fuel types, forest and grass. Sensitivity was evaluated in terms of the gross area burnt when the input of interest was systematically changed while all other inputs were held constant. For all evaluations except relative wind direction, model sensitivities were compared to an equivalent area burnt using the corresponding McArthur Forest or grassland fire danger meter (assuming an elliptical fire shape). The combination of wind direction and slope resulted in simulated fires that were not elliptical, so comparisons with shapes generated with the fire danger meters were not valid. PHOENIX predictions differed from those generated using point estimates for some circumstances; however without further investigation it is unclear on what is causing these differences. Differences in predictive performance are not necessarily representative of model error, as there are a number of differing assumptions between the systems used. However, specific situations have been flagged for follow up work. Two case study areas were used for PHOENIX model sensitivity evaluation; Wangary and Kilmore. Case study fires were simulated using observations from the day that the fires occurred with one input systematically varied. Three inputs were evaluated using the case studies; simulation resolution, start time (simulating fire ignition to occur earlier and later than observed) and start location (varying the ignition location in space). Sensitivity was evaluated by considering the change in the Area Difference Index (ADI, an index of the ratio between incorrectly predicted burnt area and the correctly predicted burnt area) from the baseline scenario (simulation resolution of 180m, ignition location and time as observed. Predictive performance varied wide with changing inputs. In general as the difference in input value to the {\textquoteleft}best estimate{\textquoteright} increased, predictive performance degraded.}, author = {Chong, Derek and Tolhurst, K.G. and Thomas Duff and Brett Cirulis} } @article {BF-4292, title = {Evaluation of weather data at different spatial and temporal scales on fire behaviour prediction using PHOENIX RapidFire 4.0 - Kilmore Case Study}, year = {2012}, abstract = {High resolution models might be expected to produce more accurate predictions, but in the case of weather forecasting data used to predict the spread of the 2009 Kilmore fire, this was not found to be true. Current weather forecasts are available on a 3600 m grid at hourly intervals. In time, computing power will enable finer spatial and temporal forecast weather to be produced operationally. This study was undertaken to understand what benefit finer scaled data might be to fire spread prediction. PHOENIX RapidFire was used to model the 2009 Kilmore fire with different spatial and temporal weather inputs. Because the fire of this size interacts with the local weather, it was found that courser level weather inputs performed better than very fine resolution data. Overall, weather forecasts at 30 minutes intervals and 1200 m spacing provided the best inputs for matching the progression of the fire. Once the fire had reached about 100,000 ha, 60 minute, 3600 m data gave the best predictions. Modelling weather at 400 m resolutions and at 5 minute intervals provides good insights into the dynamics of the weather which assists a weather forecaster, but that additional detail is not of the same benefit to fire spread predictions because large fires "smooth" the weather, terrain and fuel in the landscape. More case-studies need to be undertaken, including smaller fires burning under milder conditions to better understand the relationship between weather data scale and fire spread prediction.}, author = {Chong, Derek and Thomas Duff and Tolhurst, K.G.} } @article {BF-4291, title = {Incorporating Vertical Winds into PHOENIX RapidFire{\textquoteright}s Ember Dispersal Model}, year = {2012}, month = {12/2012}, abstract = {PHOENIX RapidFire is a fire simulation model developed in Australia as part of the Bushfire CRC. One unique aspect of the modelling process is the way spotfires are modelled and incorporated into fire spread. Part of the spotting process requires modelling of ember release, transport and spotfire ignition. In the first iteration of spotfire modelling, surface wind (10 m in the open) was used to estimate the distance and spread of embers falling ahead of a fire. However, it is acknowledged that upper-winds usually differ in speed and direction from surface winds. It was therefore thought that that fire modelling could be improved by using the wind speed and direction data produced by the numerical weather prediction models. PHOENIX was modified to incorporate multi-level wind data, however, the forecast data had an uncorrected bias that results in incorrect ember transport modelling results. The lack of systematic, high resolution and comprehensive upper-wind observations makes bias correction and fire model testing impossible at the current time. It was therefore not possible to develop an improved ember transport model for PHOENIX. This report describes the process and results obtained from this research. In the end, a version of PHOENIX that can use a single layer of upper-wind data as input was developed. This will allow some theoretical testing of PHOENIX, but cannot be used for operational fire predictions with the currently available data.}, author = {Chong, Derek and Tolhurst, K.G. and Thomas Duff} } @article {BF-4294, title = {PHOENIX RapidFire 4.0 Convection and Ember Dispersal Model}, year = {2012}, month = {12/2012}, abstract = { PHOENIX RapidFire is unique in the way it models and incorporates the spotting process in bushfires. There are three main components to the spotting process modelled: lofting, transport, and spotfire ignition. The method of modelling these three processes is described. Lofting is related to the convective strength of the fire and the amount of available ember material, transport is related to wind speed and direction, and spotfire ignition is related to the available fine fuel on the ground and its moisture content. An absence of reliable ember transport and spotfire data has necessitated an empirical approach using the Black Saturday fires in Victoria for model development. As a result, there may be some aspects of the spotting process not well captured in PHOENIX, but to date, the results on new fires have been encouraging. The effective rate of spread of bushfires can be a factor of two or three times greater if the spotting process is well developed compared with a fire, burning under generally similar conditions, but without spotting contributing to the effective rate of spread. Being able to effectively model the spotting process has dramatically increased our ability to model bushfires in eucalypt forests. Modelling the conditions associated with the spotting process has also improved our ability to model potential house loss.}, author = {Chong, Derek and Tolhurst, K.G. and Thomas Duff} } @article {BF-4293, title = {PHOENIX RapidFire 4.0{\textquoteright}s Convective Plume Model}, year = {2012}, month = {12/2012}, abstract = {Bushfires are a 3-dimensional phenomenon with significant interaction between the surface and the atmosphere. Complex coupled fire-atmosphere models have been produced and give amazingly realistic results, however the computational complexity means that they take many times real-time to run, even on super computers, and are therefore restricted to small areas and short periods of time. PHOENIX RapidFire is primarily a 2-dimensional fire model and only takes a few minutes to run fires in excess of 100,000 ha. This report describes how PHOENIX RapidFire has been developed to include elements of plume development and ember transport resulting in spot fires. This was done to try and capture some of the important 3-dimensional aspects of bushfires without large computational overheads. Development of the plume rise and spotting components of PHOENIX has been done with the knowledge of some of the key thermodynamic processes, but a number of assumptions have been made. Validation of the plume rise and spotting model is difficult because there are no detailed observations recorded for the plume, embers, spotfires and upper-level winds. Ground-based weather radar data was found to be a useful validation dataset for the plume model. The plume model in PHOENIX was calibrated against weather radar data recorded on Black Saturday, 2009. Early indications are that there has been a significant improvement in the simulation of the Black Saturday fires with the incorporation of the plume and spotting models. Further testing will be required to fully understand the limitations of the model.}, author = {Chong, Derek and Tolhurst, K.G. and Thomas Duff} } @article {BF-3174, title = {Procrustes based metrics for spatial validation and calibration of two-dimensional perimeter spread models: A case study considering fire}, journal = {Agricultural and Forest Meteorology}, volume = {160}, year = {2012}, month = {7/2012}, pages = {110 - 117}, abstract = {A number of phenomena in natural systems exhibit spread from a point source facilitated by a transport vector. Such occurrences are an important focus of landscape management, and include fires, wind driven disease and pollutant spills. Two-dimensional dynamic spread models are used to simulate the impacts of such events, determine risks and optimise responses. These models produce spatially coherent outputs that are not easily verified through traditional regression approaches. Validation of predictions is an essential part of model development and is necessary for the improvement of predictive performance. Current methods of evaluation are rarely systematic and are typically undertaken through subjective comparison of simulation outputs with observed features. There are few methods suitable for the objective analysis of freeform spread patterns, and it is proposed that a pseudo-landmark approach be adopted to allow the use of landmark based analysis methods. Vector driven spread patterns exhibit a degree of spatial structure, with distinct origin points and elongate shapes resulting from the predominant vector trajectory. These can be used as references to generate analogous landmarks for perimeter comparison. To describe differences, three indices derived from Procrustes analysis are proposed. These provide metrics to evaluate differences in perimeter orientation, size and shape. A case study simulating wildfire spread was used to demonstrate the proposed methodology. It was found to be effective for the description of perimeter differences and has potential for the validation and calibration of spread models. A number of assumptions were recognised and limitations in assigning pseudo-landmarks considered.}, issn = {01681923}, doi = {10.1016/j.agrformet.2012.03.002}, author = {Thomas Duff and Chong, Derek and Peter Taylor and Tolhurst, K.G.} }